Overview
This study focuses on the development of two new orthogonal multiwavelets possessing supercompact support. The construction methodology involves the application of Fast Bauer's method for matrix spectral factorization to the matrix product filter of the orthogonal CL multiwavelet filter. These newly derived multiwavelets exhibit characteristics of orthogonality and symmetry/antisymmetry. One of the new multiwavelets is noted for providing enhanced coding and smoothness compared to other supercompact multiwavelets.
Research Context
Multiwavelets are a generalization of wavelets, capable of providing multiple scaling functions and wavelets. Orthogonal multiwavelets maintain the properties of orthogonality, which can be beneficial in various signal processing applications. The concept of supercompact support indicates that the multiwavelet functions have a very limited spatial extent, which can be advantageous for computational efficiency and localized analysis.
Approach
The construction of the new orthogonal multiwavelets leveraged Fast Bauer's method. This method was specifically applied to perform matrix spectral factorization on the matrix product filter of an existing orthogonal multiwavelet, the CL multiwavelet filter. This systematic approach facilitated the derivation of two distinct multiwavelet filters.
Findings
The newly constructed multiwavelets demonstrate properties of orthogonality and symmetry/antisymmetry. A specific observation is that one of these new multiwavelets offers superior coding and smoothness characteristics when compared against other multiwavelets categorized as supercompact.
Performance evaluation involved a comparative analysis across several applications, including subband-based edge detection, grayscale image compression, color image compression, 1D signal denoising, and 2D signal denoising. The new multiwavelet filters were benchmarked against established multifilters: GHM, SA4, CL, Integer Haar, and Alpert. The comparative analysis indicated that the new multiwavelets can provide improved human visual measures, SSIM (Structural Similarity Index Measure), and MS-SSIM (Multi-scale Structural Similarity Index Measure) in image compression and denoising applications.
Why This Matters
The development of multiwavelets with improved performance metrics, such as better human visual measures, SSIM, and MS-SSIM in image compression and denoising, suggests potential for enhanced quality in media processing applications. Their supercompact support and inherent properties like orthogonality and symmetry/antisymmetry could offer efficiencies in computational tasks related to signal analysis and synthesis.